[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M79DH0RZ3Q044NWN7GS2JPRT":3,"skill-same:01M79DH0RZ3Q044NWN7GS2JPRT":52},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":14,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":29,"ownerDescription":30,"updatedAt":31,"install":32,"resources":45},"01M79DH0RZ3Q044NWN7GS2JPRT","coding-orchestration",null,"Coordinate bounded repository work with Astra decisions, OpenCode Go execution, risk-proportionate review and explicit provider fallback.","AO-HyS\u002Fdevelopment-system","AO-HyS","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F286554351?v=4","01M79DCXZGC6076EH5QATNXCY2","AO HyS","01M79DCY06TSBCJ946VP0BT0RG",0,"MIT","https:\u002F\u002Fgithub.com\u002FAO-HyS\u002Fdevelopment-system\u002Fblob\u002Fmain\u002Fartifacts\u002F1.11.0\u002Fskills\u002Finternal\u002Fcoding-orchestration\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: coding-orchestration\ndescription: Coordinate bounded repository work with Astra decisions, OpenCode Go execution, risk-proportionate review and explicit provider fallback.\n---\n\nFor visual work, finish Impeccable, obtain independent Astra visual critique,\nresolve its findings, then record final delivery evidence from that candidate.\nWorking screenshots may support critique. Astra retains live Computer Use;\nLuna may prepare bounded scripts or media files, never inherit browser control\nmerely to reduce model cost. See `design-quality` and `evidence-capture`.\n\n\nThe parent owns the full requested result, architecture, integration, conflicts\nand verification. Read `$HOME\u002F.codex\u002Fdevelopment-system\u002Fagent-roster.json` before\nselecting a model. Keep the task contract concise. Ordinary bounded work uses\none execution and parent review; use `orchestration-plan` when ownership,\nobserved risks or parallel dependencies need an explicit plan. The current user's\nauthorization governs dispatch; a plan is not an authorization receipt.\n\n## Execution\n\nAstra orchestrates, designs and judges acceptance. Use `evidence-capture` for\nrecordings and walkthroughs; deterministic host tools run stable batches.\nThe evidence-preparation route selects Luna for bounded capture scripts and\nmedia preparation. Astra drives novel or ambiguous UI states when needed.\nUse local tools directly for deterministic search, commands and collection.\nBatch independent calls and overlap independent CI with useful work. Serialize\nmutating commands on the same writer\u002Fbuild state. Delegation must shorten the\ncritical path or add needed independent judgment; do not create handoffs merely\nbecause tools support them.\nDelegate bounded implementation to the ordered OpenCode Go-first fast route\nwhen a worker reduces delivery time. Factory\u002FDevin provide declared fallbacks\nand Fable help with complex review. Luna is the final worker fallback.\n\nEach worker receives the exact root\u002Frevision, owned files, behavior, constraints,\nfocused checks, expected artifact and stop condition. Verify path confinement\nbefore writable dispatch. Keep one writer per surface; independent writers may\nrun concurrently only with explicit disjoint ownership and integration owned\nby Astra. Preserve concurrent work. Reuse the worker when its context is useful.\n\nUse the provider's real CLI argument array from `model-route`. OpenCode Go is\n`opencode run --pure --model opencode-go\u002F\u003Cmodel> --variant \u003Ceffort> --format json`.\nNever add permission or publication authority inside the pure resolver. Record\nthe provider's observed model before claiming resolution. Cache typed quota or\navailability failures with `record-provider-failure --input \u003Cobservation.json>`\nand bounded expiry (at most seven days). `model-route` reads that private host\ncache unless an explicit observation packet overrides it. Do not retry an\nunexpired quota failure on each task; carry useful work into the next route.\nOn T3, launch CLI workers through `development-system run-worker --input\npacket.json --json`; read `worker-reference.md` for the packet. Its deterministic\nowner watch stops the process group when the turn ends or becomes unreadable.\nOther hosts must keep workers attached to their native cancellation lifecycle.\nCheck startup, then completion or an observed stall; avoid repeated model-driven\npolling. A productive worker\ndoes not fail merely because a fixed number of minutes passed.\n\n## Review and evidence\n\nTreat simplification, test-value review, correction and objective verification\nas responsibilities, not mandatory new conversations. Astra reviews a normal\nworker's diff and the objective. Request Fable for observed complex risk or an\nexplicit independent-review request. Review corrections and remaining risks;\ndo not restart the whole repository analysis. Structured review is opt-in\n(`signals.structuredReview: true`) or justified by observed risk. Preserve\nhost-bound correction validation and never close an\nunresolved blocker\u002Fhigh finding because a reviewer timed out.\n\nRun focused checks after edits and applicable integration checks once when the\ncandidate stabilizes. Do not create tests that copy implementation details or\nrepeat broad suites without a concrete requirement. File\u002Ftest counts and\nstyling scores are not quality gates. Verify the public behavior being claimed. After affected checks and required\ngates pass, continue toward the authorized delivery. Repeat only for relevant\nchanges, failures or unresolved concerns; reuse evidence only while its tested\ncandidate, inputs and environment still apply. Never bypass required hooks\u002FCI.\nPreserve the check exit code: run the check into a log, then read relevant output\nseparately. A successful trailing output filter can hide a failed test.\n\nUse `design-quality` with Impeccable when visual quality is acceptance. An independent Astra\nfinish reviewer satisfies the critique responsibility; the designer reviewing\nits own work does not. Reuse that review instead of duplicating it. Use one focused Jakub skill\nfor the detail being improved. `better-interface` is for a broad interface\nreview; do not force it onto every UI change. Prefer existing product tokens\nand conventions over generic polish defaults. Show-me or PR Lens can explain\na complex change; ordinary diffs do not require visual artifacts.\n\nComputer Use is opt-in to acceptance and led by Astra. A delegated neutral\nexecutor gets only a SHA-256-bound execution plan, authorized origins\u002Fpaths,\nactions and opaque inputs; keep the private rubric with the parent. The host\nvalidates write authority separately. Capture relevant evidence early and keep\nruntime\u002Ftool failures distinct from product failures. Recording itself does\nnot require another model. Avoid one inference per stable click or frame.\nOne driver owns each session; exchange completed evidence instead of progress\nchatter. Recover readiness before retrying a failed capture. Respect the repository's allowed browser mechanism.\n\nReport the delivered outcome, checks, remaining limitations and exact release\nstate. Retain requested\u002Fresolved models, fallback causes and lane terminal\nstates in the evidence. File installation alone does not prove live influence.\n\nChoose the smallest useful change for user experience, developer workflow and\nagent comprehension (UX, DX, AX). These are decision criteria, not three new\naudit stages. Request Astra Low for clear orchestration, Medium for ambiguity\nand High for difficult judgment. Verify active host effort before claiming it\nchanged; the roster alone cannot change an already-running turn.\n","For visual work, finish Impeccable, obtain independent Astra visual critique,\nresolve its findings, then record final delivery evidence from that candidate.\nWorking screenshots may support critique. Astra retains live Computer Use;\nLuna may prepare bounded scripts or media files, never inherit browser control\nmerely to reduce model cost. See `design-quality` and `evidence-capture`.\n\n\nThe parent owns the full requested result, architecture, integration, conflicts\nand verification. Read `$HOME\u002F.codex\u002Fdevelopment-system\u002Fagent-roster.json` before\nselecting a model. Keep the task contract concise. Ordinary bounded work uses\none execution and parent review; use `orchestration-plan` when ownership,\nobserved risks or parallel dependencies need an explicit plan. The current user's\nauthorization governs dispatch; a plan is not an authorization receipt.\n\n## Execution\n\nAstra orchestrates, designs and judges acceptance. Use `evidence-capture` for\nrecordings and walkthroughs; deterministic host tools run stable batches.\nThe evidence-preparation route selects Luna for bounded capture scripts and\nmedia preparation. Astra drives novel or ambiguous UI states when needed.\nUse local tools directly for deterministic search, commands and collection.\nBatch independent calls and overlap independent CI with useful work. Serialize\nmutating commands on the same writer\u002Fbuild state. Delegation must shorten the\ncritical path or add needed independent judgment; do not create handoffs merely\nbecause tools support them.\nDelegate bounded implementation to the ordered OpenCode Go-first fast route\nwhen a worker reduces delivery time. Factory\u002FDevin provide declared fallbacks\nand Fable help with complex review. Luna is the final worker fallback.\n\nEach worker receives the exact root\u002Frevision, owned files, behavior, constraints,\nfocused checks, expected artifact and stop condition. Verify path confinement\nbefore writable dispatch. Keep one writer per surface; independent writers may\nrun concurrently only with explicit disjoint ownership and integration owned\nby Astra. Preserve concurrent work. Reuse the worker when its context is useful.\n\nUse the provider's real CLI argument array from `model-route`. OpenCode Go is\n`opencode run --pure --model opencode-go\u002F\u003Cmodel> --variant \u003Ceffort> --format json`.\nNever add permission or publication authority inside the pure resolver. Record\nthe provider's observed model before claiming resolution. Cache typed quota or\navailability failures with `record-provider-failure --input \u003Cobservation.json>`\nand bounded expiry (at most seven days). `model-route` reads that private host\ncache unless an explicit observation packet overrides it. Do not retry an\nunexpired quota failure on each task; carry useful work into the next route.\nOn T3, launch CLI workers through `development-system run-worker --input\npacket.json --json`; read `worker-reference.md` for the packet. Its deterministic\nowner watch stops the process group when the turn ends or becomes unreadable.\nOther hosts must keep workers attached to their native cancellation lifecycle.\nCheck startup, then completion or an observed stall; avoid repeated model-driven\npolling. A productive worker\ndoes not fail merely because a fixed number of minutes passed.\n\n## Review and evidence\n\nTreat simplification, test-value review, correction and objective verification\nas responsibilities, not mandatory new conversations. Astra reviews a normal\nworker's diff and the objective. Request Fable for observed complex risk or an\nexplicit independent-review request. Review corrections and remaining risks;\ndo not restart the whole repository analysis. Structured review is opt-in\n(`signals.structuredReview: true`) or justified by observed risk. Preserve\nhost-bound correction validation and never close an\nunresolved blocker\u002Fhigh finding because a reviewer timed out.\n\nRun focused checks after edits and applicable integration checks once when the\ncandidate stabilizes. Do not create tests that copy implementation details or\nrepeat broad suites without a concrete requirement. File\u002Ftest counts and\nstyling scores are not quality gates. Verify the public behavior being claimed. After affected checks and required\ngates pass, continue toward the authorized delivery. Repeat only for relevant\nchanges, failures or unresolved concerns; reuse evidence only while its tested\ncandidate, inputs and environment still apply. Never bypass required hooks\u002FCI.\nPreserve the check exit code: run the check into a log, then read relevant output\nseparately. A successful trailing output filter can hide a failed test.\n\nUse `design-quality` with Impeccable when visual quality is acceptance. An independent Astra\nfinish reviewer satisfies the critique responsibility; the designer reviewing\nits own work does not. Reuse that review instead of duplicating it. Use one focused Jakub skill\nfor the detail being improved. `better-interface` is for a broad interface\nreview; do not force it onto every UI change. Prefer existing product tokens\nand conventions over generic polish defaults. Show-me or PR Lens can explain\na complex change; ordinary diffs do not require visual artifacts.\n\nComputer Use is opt-in to acceptance and led by Astra. A delegated neutral\nexecutor gets only a SHA-256-bound execution plan, authorized origins\u002Fpaths,\nactions and opaque inputs; keep the private rubric with the parent. The host\nvalidates write authority separately. Capture relevant evidence early and keep\nruntime\u002Ftool failures distinct from product failures. Recording itself does\nnot require another model. Avoid one inference per stable click or frame.\nOne driver owns each session; exchange completed evidence instead of progress\nchatter. Recover readiness before retrying a failed capture. Respect the repository's allowed browser mechanism.\n\nReport the delivered outcome, checks, remaining limitations and exact release\nstate. Retain requested\u002Fresolved models, fallback causes and lane terminal\nstates in the evidence. File installation alone does not prove live influence.\n\nChoose the smallest useful change for user experience, developer workflow and\nagent comprehension (UX, DX, AX). These are decision criteria, not three new\naudit stages. Request Astra Low for clear orchestration, Medium for ambiguity\nand High for difficult judgment. Verify active host effort before claiming it\nchanged; the roster alone cannot change an already-running turn.\n",{"name":5,"description":7},"main","9f5b43aadb1a92441d65e9cb74a584cd6fca62bc","artifacts\u002F1.11.0\u002Fskills\u002Finternal\u002Fcoding-orchestration","https:\u002F\u002Fgithub.com\u002FAO-HyS\u002Fdevelopment-system","https:\u002F\u002Fgithub.com\u002FAO-HyS","https:\u002F\u002Faohys.net\u002F","Software Boutique, we make your projects come true live. ","2026-11-14 03:17:27",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add AO-HyS\u002Fdevelopment-system --skill coding-orchestration",{"tool":38,"command":39},"bunx","bunx skills add AO-HyS\u002Fdevelopment-system --skill coding-orchestration",{"tool":41,"command":42},"pnpm","pnpm dlx skills add AO-HyS\u002Fdevelopment-system --skill coding-orchestration","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FAO-HyS\u002Fdevelopment-system\u002Fblob\u002Fmain\u002Fartifacts\u002F1.11.0\u002Fskills\u002Finternal\u002Fcoding-orchestration\u002FSKILL.md",true,[46],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"worker-reference.md","REFERENCE","参考文档",2075,"https:\u002F\u002Fgithub.com\u002FAO-HyS\u002Fdevelopment-system\u002Fblob\u002Fmain\u002Fartifacts\u002F1.11.0\u002Fskills\u002Finternal\u002Fcoding-orchestration\u002Fworker-reference.md",[53,60,67,74,81,88],{"skillNo":54,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":57,"categories":58,"keywords":59,"publishedAt":19},"01M79EQ4BPYQTX8RWN1Y1J2K1F","working-backwards","Define a software feature through PRFAQ and staged documents when the user requests working backwards before coding.","https:\u002F\u002Fgithub.com\u002FAO-HyS\u002Fdevelopment-system\u002Fblob\u002Fmain\u002Fartifacts\u002F2.8.0\u002Fskills\u002Finternal\u002Fworking-backwards\u002FSKILL.md",[],[],{"skillNo":61,"name":62,"nameZh":6,"description":63,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":64,"categories":65,"keywords":66,"publishedAt":19},"01M79EQ2E8P85Y1Y7VNJZMNEJE","flow-implement","Implement a requested software change through verification, corrections, and its authorized delivery endpoint.","https:\u002F\u002Fgithub.com\u002FAO-HyS\u002Fdevelopment-system\u002Fblob\u002Fmain\u002Fartifacts\u002F2.8.0\u002Fskills\u002Finternal\u002Fflow-implement\u002FSKILL.md",[],[],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":19},"01M79EQ0PNMKHPJ78F8WMNX06P","development-steward","Run the scheduled, read-only weekly maintenance review for the six explicitly opted-in repositories and prepare one concise private report for Check-in.","https:\u002F\u002Fgithub.com\u002FAO-HyS\u002Fdevelopment-system\u002Fblob\u002Fmain\u002Fartifacts\u002F2.8.0\u002Fskills\u002Finternal\u002Fdevelopment-steward\u002FSKILL.md",[],[],{"skillNo":75,"name":76,"nameZh":6,"description":77,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":80,"publishedAt":19},"01M79EPXJ56MX7955GWP1HD4B0","grill-with-docs","Run one recoverable interview for functional, visual, or mixed software grills; reuse equivalent answers and offer the shared HTML questionnaire.","https:\u002F\u002Fgithub.com\u002FAO-HyS\u002Fdevelopment-system\u002Fblob\u002Fmain\u002Fartifacts\u002F2.7.0\u002Fskills\u002Finternal\u002Fgrill-with-docs\u002FSKILL.md",[],[],{"skillNo":82,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":85,"categories":86,"keywords":87,"publishedAt":19},"01M79EPTFPHYEV40ZG52Z0GVS2","global-agent-guardrails","Install, audit, test, or update the Development System's global destructive-command guard for Codex, Claude Code and T3 Code. Use when configuring agent safety hooks or investigating whether dangerous shell operations are blocked.","https:\u002F\u002Fgithub.com\u002FAO-HyS\u002Fdevelopment-system\u002Fblob\u002Fmain\u002Fartifacts\u002F2.7.0\u002Fskills\u002Finternal\u002Fglobal-agent-guardrails\u002FSKILL.md",[],[],{"skillNo":89,"name":90,"nameZh":6,"description":91,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":92,"licenseFileUrl":93,"sourceHtmlUrl":94,"categories":95,"keywords":96,"publishedAt":19},"01M79EPR0KPHKAFN7V6CEEDR0Y","impeccable","Design or refine frontend UI using project context and a focused design playbook; also handles visual critique and polish.","Apache-2.0","https:\u002F\u002Fgithub.com\u002FAO-HyS\u002Fdevelopment-system\u002Fblob\u002Fmain\u002Fartifacts\u002F2.7.0\u002Fskills\u002Fexternal\u002Fimpeccable\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FAO-HyS\u002Fdevelopment-system\u002Fblob\u002Fmain\u002Fartifacts\u002F2.7.0\u002Fskills\u002Fexternal\u002Fimpeccable\u002FSKILL.md",[],[]]